Mathai Mammen, Global Head of R&D at Janssen Pharmaceuticals, presented at MIT AIDM Conference in February earlier this year, giving an insight into how data science is disrupting biopharma R&D.
Mammen begins by looking at recent examples of first-in-class products and fundamental changes taking place in medicine and drug development. These changes are all the more impressive given the innate improbability of drug development succeeding. He continues to explain the war we’re currently engaged in against bias as we try to merge technology and life sciences. These historical biases still determine how we approach many projects today.
In relation to drug discovery, Mammen outlines how data science is influencing how we generate high value biological insights and validate drug targets. We have enough human data, be it from genetics, phenotypic data, and sythetic organoids, to now use human systems to predict human disease. It’s still a relatively novel approach for us, to not rely so heavily on animal models to study disease. AI and ML predictions in drug discovery phases can reduce the time and improve the quality of new molecular entities (NMEs).
Other points Mammen covers is the scarcity of bilingual individuals i.e. those that have both the data science AND life sciences domain expertise, within pharma. He recommends going out of your way to persue bilinguality, or as close as you can manage in both domains to be at a real professional advantage. At Janssen, his number one most important objective for the company is the full incorporation of data science into every element of what they do.
Within the very brief Q&A time Mammen addresses the following audience questions:
- As collecting data is both a software engineering and a social engineering problem. What are some ways in which you have convinced your scientists that collecting data in a systematic fashion will help their work in the long run? (27:23)
- How does data science rank in terms of priorities at Janssen? (29:10)
- What has been the biggest impact of AI/ML at Janssen? (29:29)
- How much deep learning do you use at Janssen (31:18)